While the speech synthesis is running the tab doesn't get muted. Steps to reproduce: 1. Ensure that the speech synthesis is enabled (media.webspeech.synth.enabled must be true) 2. Go to http://codepen.io/matt-west/pen/wGzuJ 3. Type "This is some text to test the speech synthesis of Firefox." 4. Click the 'Speak' button 5. Why the synthesis is running click the mute icon within the tab or press Ctrl+M => The synthesis continues even when the tab is muted. (Not sure whether this only happens on Windows or it is an issue on all platforms.) Sebastian

When stating speech on muted tab, it is bug 1225347. But when set muted during specking, it is limitation of indirect speech service. Example, On SAPI, changing volume will be applied per word boundary. If changing Windows backend to direct speech service, we can improve it.

(In reply to Sebastian Zartner [:sebo] from comment #0) > 5. Why the synthesis is running click the mute icon within the tab or press Ctrl+M This step should of course say "While the synthesis ..." (In reply to Makoto Kato [:m_kato] from comment #1) > When stating speech on muted tab, it is bug 1225347. But when set muted > during specking, it is limitation of indirect speech service. As you can see in step 5 this happens when you mute it during speaking. > Example, On SAPI, changing volume will be applied per word boundary. You're right. It is now muted on word boundary. Before I filed the bug it continued speaking until the end of the entered phrase. So it already improved meanwhile. > If changing Windows backend to direct speech service, we can improve it. Good to know. Maybe that should be reported to Microsoft? Sebastian

(In reply to Sebastian Zartner [:sebo] from comment #2) > As you can see in step 5 this happens when you mute it during speaking. We call OS API to change volume, but it depends on Speech engine. > > If changing Windows backend to direct speech service, we can improve it. > > Good to know. Maybe that should be reported to Microsoft? Actually, our speech service has 2 types, direct audio and indirect audio. Indirect audio service manages the audio by OS mixer. So it depends on speech engine and OS implementation. But, since SAPI can get raw audio stream that we can manage on our mixer, to resolve this issue, we have to change to direct audio service.